Deteksi Lesi Periodontitis Apikal berdasarkan Skala Periapical Index (PAI) menggunakan Citra Radiografi Panoramik dengan Metode EfficientDet

Abstract

Diagnosing apical periodontitis lesions via panoramic radiography faces significant challenges, including suboptimal image quality, digital noise, blurred image boundaries, and overlapping anatomical structures. These conditions not only complicate the diagnostic process and increase the demand on dentists' time and effort, but also potentially lead to inconsistent assessments and an increased risk of misdiagnosis due to interpretations highly dependent on individual clinicians' skills and experience. To address these challenges, this study developed an artificial intelligence-based system using the EfficientDet method to detect apical periodontitis lesions based on the Periapical Index (PAI) scale. Lesion were classified into three categories: PAI 3, PAI 4, and PAI 5. Based on test results, the model utilizing the EfficientDet algorithm was able to detect apical periodontitis lesions in panoramic radiographic images with an accuracy rate of 89.95%. These results indicate that the model performs well in recognizing apical periodontitis lesions and determining their severity according to the Periapical Index Scale.